Machine Learning Engineer - ML Training Platform
Summary
Build and optimize a distributed ML training substrate that trains large models across many low-bandwidth nodes using model parallelism and P2P networking.
Job Description
\n Overview\n
Pluralis Research carries out foundational research on Protocol Learning: multi-participant training of foundation models where no single participant has, or can ever obtain, a full copy of the model. The purpose of Protocol Learning is to facilitate the creation of community-trained and community-owned frontier models with self-sustaining economics.
\nWe're looking for Senior/Staff engineers with 5+ years of experience in distributed systems and ML large-scale training. You'll be implementing a novel substrate for training distributed ML models that work under consumer grade internet connection.
Responsibilities Distributed Training Architecture & Optimization\n- \n
- \n
Design and implement large-scale distributed training systems optimized for heterogeneous hardware operating under low-bandwidth, high-latency conditions.
\n - \n
Develop and optimize model-parallel training strategies (data, tensor, pipeline parallelism) with custom sharding techniques that minimize communication overhead.
\n - \n
Optimize GPU utilization, memory efficiency, and compute performance across distributed nodes.
\n - \n
Implement robust checkpointing, state synchronization, and recovery mechanisms for long-running, fault-prone training jobs.
\n - \n
Build monitoring and metrics systems to track training progress, model quality, and system bottlenecks.
\n
- \n
- \n
Architect resilient training systems where nodes can fail, networks can partition, and participants can dynamically join or leave.
\n - \n
Design and optimize peer-to-peer topologies for decentralized coordination across non-co-located nodes.
\n - \n
Implement NAT traversal, peer discovery, dynamic routing, and connection lifecycle management.
\n - \n
Profile and optimize communication patterns to reduce latency and bandwidth overhead in multi-participant environments.
\n
- \n
- \n
Strong experience building and operating distributed systems in production.
\n - \n
Hands‑on expertise with distributed training frameworks (FSDP, DeepSpeed, Megatron, or similar).
\n - \n
Deep understanding of model parallelism (data, tensor, pipeline parallelism).
\n - \n
Expert‑level Python with production experience (concurrency, error handling, retry logic, clean architecture).
\n - \n
Strong networking fundamentals: P2P systems, gRPC, routing, NAT traversal, distributed coordination.
\n - \n
Experience optimizing GPU workloads, memory management, and large‑scale compute efficiency.
\n
- \n
- \n
Equity‑heavy compensation with meaningful ownership in a mission‑driven company
\n - \n
Competitive base salary for senior engineering roles in Australia
\n - \n
Visa sponsorship available for exceptional candidates
\n - \n
Remote‑first with optional access to our Melbourne hub
\n - \n
World‑class team — teammates were previously at Google, Amazon, Microsoft, and leading startups
\n
Backed by Union Square Ventures and other tier‑1 investors, we're a world‑class, deeply technical team of ML researchers and engineers. Pluralis is unapologetically ideological. We view the world as a better place if we are able to implement what we are attempting, and Protocol Learning as the only plausible approach to preventing a handful of massive corporations monopolising model development, access and release, and achieving massive economic capture. If this resonates, please apply.
\n #J-18808-Ljbffr